Investigating temporal and prosodic markers in clinical high-risk for psychosis participants using automated acoustic analysis

Bianciardi, Bianca (2022) Investigating temporal and prosodic markers in clinical high-risk for psychosis participants using automated acoustic analysis. MSc(R) thesis, University of Glasgow.

Full text available as:
[img] PDF
Download (2MB)


Introduction: Research into language abnormalities has gained attention given the role of language impairments as a plausible marker for early detection and diagnosis of psychosis. Semantic and syntactic aberrations have been widely observed in schizophrenia across illness stages. Recently, acoustic abnormalities such as temporal and prosodic features of speech have been observed in schizophrenia patients. Yet, mixed evidence exists on the presence of acoustic deficits in participants meeting clinical high-risk for psychosis (CHR-P) criteria. The present study aimed to clarify whether acoustic impairments could be used to identify CHR-P individuals when compared to participants with substance use and affective disorders (clinical high-risk negative; (CHR-N) and to healthy controls (HC) participants. Crucially, methodological issues were addressed including the duration of speech samples to determine their impact on the acoustic results.

Methods: Data were available from the Youth mental health, risk and Resilience (YouR) study. Speech samples were recorded from the semi structured clinical interviews of the Comprehensive Assessment of At Risk Mental States (CAARMS) in 50 CHR-P participants who were compared against a group of 17 HC and 23 CHR-N participants. Temporal and prosodic features were extracted from the recordings. Linear regression was used to determine the influence of interview duration on the acoustic estimates. After examining group differences for each of the acoustic features, temporal and prosodic indices were used to determine whether they could be used determine group status using binary logistic regressions.

Results: No deficits were observed in temporal or prosodic variables in the CHR-P group when compared to HCs. Instead, CHR-N individuals were characterized by slower speech rate, more and longer pauses and higher unvoiced frames percentage compared to CHR-P participants. Temporal features could better discriminate between groups compared to prosodic features, with models explaining up to 47% of the variance between CHR-Ns and HCs and up to 28% of variance between CHR-Ps and CHR-Ns. Yet, none of these models survived bootstrapping. Moreover, group differences for temporal and prosodic features were largely robust to the interview duration effects. Finally, no significant relationship was obtained for temporal and prosodic features with clinical and functional symptom severity.

Discussion: These finding suggests that temporal and prosodic features of speech are not impaired in early-stage psychosis. The acoustic features examined indicated the presence of acoustic impairments in CHR-N participants, which resulted spurious following bootstrapping and therefore hinted to the importance of employing validation methods on acoustic signatures in psychosis. This is crucial given the small sample sizes across the literature and heterogeneity of the clinical groups. Given the absence of acoustic disturbances of speech in CHR-P individuals observed in the present research, sematic and syntactic abnormalities may constitute a more promising biomarker of early psychosis. Further studies are required to clarify whether acoustic abnormalities are present in sub-groups of CHR-P participants with elevated psychosis-risk.

Item Type: Thesis (MSc(R))
Qualification Level: Masters
Subjects: B Philosophy. Psychology. Religion > BF Psychology
Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Psychology & Neuroscience
Supervisor's Name: Fracasso, Dr. Alessio and Rousselet, Dr. Guillaume A.
Date of Award: 2022
Depositing User: Theses Team
Unique ID: glathesis:2022-83313
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 16 Jan 2023 11:54
Last Modified: 16 Jan 2023 11:54
Thesis DOI: 10.5525/gla.thesis.83313

Actions (login required)

View Item View Item


Downloads per month over past year